Which study design best supports causal inference when evaluating a school-wide math intervention deployed in only half the grade levels this year?

• A. A cross-sectional survey of all students this year.
• B. A randomized or matched quasi-experimental design with a control group and pre/post measures.
• C. A descriptive case study of a single classroom.
• D. A purely correlational analysis of test scores across all grades.

Answer: (B)

To draw causal conclusions about a school intervention, you need a design that controls for preexisting differences and shows that changes occur after the intervention. A randomized or matched quasi-experimental design with a control group and pre/post measures fits this requirement because it creates comparable groups (through randomization or careful matching) and tracks outcomes before and after the intervention. This setup lets you compare how scores change in the treated half of the grades versus the untreated half, while the pre/post data help separate the treatment effect from natural maturation or outside trends. Control of confounding factors and temporal sequencing are essential for causal inference, which purely descriptive or correlational approaches cannot provide. A cross-sectional snapshot captures one moment in time and can't reveal what caused changes; a descriptive case study of a single classroom lacks breadth and a control group; a purely correlational analysis across grades cannot rule out alternative explanations for associations between scores and the intervention.